Otimização multiobjetivo aplicada a processos logísticos de resíduos de papel: abordagens de programação por metas determinísticas e sob incerteza
Carregando...
Data
2021-06-30
Autores
Defalque, Cristiane Maria [UNESP]
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Estadual Paulista (Unesp)
Resumo
Neste estudo, foi elaborado um modelo de programação linear inteira mista multiobjetivo, multiproduto, multinível e multiperíodo para otimizar processos logísticos de centros intermediário cujos objetivos envolvem a maximização da quantidade coletada e do atendimento à demanda de fardos, a minimização da distância percorrida entre os pontos de coleta e dos custos. Para lidar com os múltiplos objetivos, foram utilizadas as abordagens de programação por metas ponderada e lexicográfica. A programação por metas multiescolha revisada também foi utilizada para incorporar incerteza na quantidade disponível para a coleta. Para a análise do modelo proposto, foram realizados testes computacionais com instâncias baseadas em dados da literatura e em dados reais de uma empresa brasileira do setor. Para todos os testes realizados, a linguagem General Algebraic Modeling System 23.6.5 e o solver CPLEX 12.2.0.2 foram utilizados para modelagem e otimização. Embora o modelo lexicográfico tenha apresentado dificuldades para encontrar soluções viáveis em alguns testes com dados reais, as abordagens determinísticas apresentaram bons resultados e mostraram que as formulações podem auxiliar o gestor no processo de tomada de decisão. Verificou-se ainda que a variação de parâmetros pode aumentar a dificuldade do modelo, o que torna difícil, muitas vezes, encontrar uma boa solução viável. Apesar disso, o modelo sob incerteza apresentou resultados competitivos em relação ao modelo determinístico, principalmente por permitir a representação de vários cenários. De uma forma geral, os resultados apontaram diferentes possibilidades de solução para auxiliar o planejamento logístico, não só sob o ponto de vista econômico, mas também ambiental e social.
In this study, a mixed multiobjective, multiproduct, multilevel, and multiperiod integer linear programming model has been developed to optimize logistics processes of intermediate centers whose objectives comprehend maximizing the collected quantity and meeting demand, minimizing the distance traveled between the collection points and costs. Weighted and lexicographic goal programmings have been used to deal with the multiple objectives. The revised multi-choice goal programming has also been used to incorporate uncertainty into the quantity available for collection. For the analysis of the proposed model, computational tests have been executed with instances based on literature data and real data of a Brazilian company in the sector. For all the performed tests, General Algebraic Modeling System 23.6.5 language and CPLEX 12.2.0.2 solver have been used for modeling and optimization. Although the lexicographic model presented difficulties to find viable solutions in some tests with real data, the deterministic approaches have presented good results and have shown that the formulations can assist managers in the decision-making process. It has also been verified that the variation of parameters can increase the difficulty of the model, which often makes it difficult to find a good viable solution. Despite this, the model under uncertainty has shown competitive results concerning the deterministic model, mainly because it allows to represent various scenarios. In general, results showed different solution possibilities to assist with logistical planning, not only from an economic point of view, but also environmental and social
In this study, a mixed multiobjective, multiproduct, multilevel, and multiperiod integer linear programming model has been developed to optimize logistics processes of intermediate centers whose objectives comprehend maximizing the collected quantity and meeting demand, minimizing the distance traveled between the collection points and costs. Weighted and lexicographic goal programmings have been used to deal with the multiple objectives. The revised multi-choice goal programming has also been used to incorporate uncertainty into the quantity available for collection. For the analysis of the proposed model, computational tests have been executed with instances based on literature data and real data of a Brazilian company in the sector. For all the performed tests, General Algebraic Modeling System 23.6.5 language and CPLEX 12.2.0.2 solver have been used for modeling and optimization. Although the lexicographic model presented difficulties to find viable solutions in some tests with real data, the deterministic approaches have presented good results and have shown that the formulations can assist managers in the decision-making process. It has also been verified that the variation of parameters can increase the difficulty of the model, which often makes it difficult to find a good viable solution. Despite this, the model under uncertainty has shown competitive results concerning the deterministic model, mainly because it allows to represent various scenarios. In general, results showed different solution possibilities to assist with logistical planning, not only from an economic point of view, but also environmental and social
Descrição
Palavras-chave
Programação por Metas Ponderada, Programação por Metas Lexicográfica, Programação por Metas Multiescolha Revisada, Reciclagem de papel, Logística Reversa, Weighted Goal Programming, Lexicographic Goal Programming, Revised Multichoice Goal Programming, Waste Paper Recycling, Reverse Logistics, Processo decisório, Programação linear